Using a self-adaptive neighborhood scheme with crowding replacement memory in genetic algorithm for multimodal optimization

In this paper a new GA based niching method using a Self-adaptive Neighborhood scheme with Crowding Replacement Memory (GA_SN_CM) for multimodal optimization is proposed, where, instead of using a niche radius to identify neighborhoods in the population, each individual attempts to select suitable neighbors from the population adaptively. Such neighborhood structure allows eliminating redundant solutions in a neighborhood to increase the diversity of the population which leads the algorithm to explore more solutions. Besides, in order to conserve found niche during the niching procedure, a memory swarm with crowding replacement scheme is used along with the main population. The results of performance comparison between the proposed method and some existing niching techniques over several multimodal benchmark functions demonstrate good performance of GA_SN_CM in improving the niching process.